MT auto-metrics for the Arabic translation layers #3

Open

opened

on Jul 23, 2026

Run automatic MT quality metrics (COMET/MetricX-style) over the frozen Layer-A translations (Gemma-3 and Gemma-4, ~31k items each). Report per-benchmark quality and correlate with observed Arabic degradation.

to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Assignees

No one assigned

Labels

No labels

No labels

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests